PXReprise

Reprocess public proteomics data at scale, for any biological question.

Documentation: https://smith-chem-wisc.github.io/PXReprise/

You describe a biological question in a small text file. PXReprise finds the relevant deposits in PRIDE, decides which it can search and how, downloads them, checks every file's quality, searches them with MetaMorpheus under one versioned method, and records exactly how every result was made. A new question is a new text file, never a change to the program.

Have an AI agent set it up

Open your coding agent (Claude Code, Codex, Cursor, ...) in an empty folder and paste:

Install PXReprise and prove it works, following https://raw.githubusercontent.com/smith-chem-wisc/PXReprise/master/AGENTS.md exactly. Ask me before accepting any licence or starting anything that runs for more than an hour.

Or do it yourself

You need the .NET 10 SDK, git, and 100 GB of free disk. No Python.

git clone https://github.com/smith-chem-wisc/PXReprise.git
cd PXReprise
dotnet build -c Release
dotnet publish src/PXReprise -c Release -o <tools>/pxreprise

Then download MetaMorpheus 1.1.11 (MetaMorpheus_CommandLine.zip), copy machines/example.toml to machines/my-machine.toml, and fill in its paths. The full walk-through, including a one-hour test run, is in Getting started.

Using it

pxreprise is one program; the first word after it says what to do:

pxreprise validate question.toml                                  # check a question file
pxreprise census question.toml --queue                            # what would be searched, and why (no downloads)
pxreprise batch run question.toml --machine machines/my-machine.toml   # do it all, deposit by deposit
pxreprise batch status question.toml                              # progress
pxreprise batch stop question.toml                                # finish the current deposit, then stop

See Commands for the rest.

How it is organised

	Answers	Lives in
Question	what is studied: PRIDE keywords, relevance rules, hand decisions	your own folder (question.toml)
Profile	how one kind of data is searched, versioned (label-free-dda@2)	profiles/
Machine	where things live on one computer	machines/my-machine.toml

PXReprise is orchestration, policy and provenance. File reading, chemistry and quantification come from mzLib; search is MetaMorpheus, run as its own process per dataset.

Developing

dotnet test --filter "Category!=ExternalService&Category!=LocalCorpus"   # offline; required
dotnet test --filter "Category=ExternalService"                         # live PRIDE/UniProt canaries; skip when a service is down

.NET 10; mzLib 1.0.593 from NuGet (MetaMorpheus 1.1.11, a separate process, uses 1.0.592). Every command prints one JSON envelope ({"ok":true,"data":...} or {"ok":false,"error":{...}}) and exits 0 (ok), 1 (handled failure) or 2 (usage).

Licence

MIT; see LICENSE. Reading Thermo .raw files requires accepting Thermo's RawFileReader licence, which is the operator's choice (accept_thermo_licence in the machine file).